Coded video: a scene behind every frame
A coded video starts with instructions for what to draw and how it should move. AI can write those instructions from your brief; a renderer turns them into frames. CodedClips brings that workflow into a browser, with 3D scenes, vector motion graphics and a render API.
Create your first coded clipTry the free vector animation sample
Sign up with no payment. Generating uses prepaid credit, with the price shown before it runs. See credit packs.
Read the render API documentation · Credits and refunds
What makes a video coded?
Instead of asking a video model to synthesize pixels directly, a model writes geometry, lights, a camera and motion, or a structured vector document. The renderer draws that scene over time. The final MP4 is an ordinary video; the difference is the source behind it.
Where it is useful
Start with geometric product concepts, abstract brand motion, simple explainers and scenes with a clear camera move. These give the model a bounded visual task. Describe the subject, materials, lighting, camera and action rather than asking for a whole film in one shot.
Try a specific brief
For a 3D shot: “A matte amber sphere rises above a dark circular pedestal. Soft side lighting, a slow camera orbit, no text, five seconds.” For vector motion: “Three cream circles assemble into a triangle on a deep green background, pause, then separate. Six seconds, gentle easing.”
Coded animation and generative AI video
A code-authored scene exposes its geometry and motion, making it useful when you want a source representation alongside the movie. Pixel-generating video models can be useful for naturalistic footage and complex visual textures. Neither approach guarantees a correct result; inspect the output before publishing.
What CodedClips delivers today
The 3D studio saves the generated Three.js scene and records a 3–20 second MP4 at 24 frames per second. The vector studio creates a validated motion document that plays in the browser. Multi-shot sequences share written direction; they do not share a persistent 3D world.
No installation or monthly subscription
Choose a model, review its quote and confirm using prepaid credits. Failed generations return app credit. Your creations stay in your account history, and you choose whether to share a link.
Questions
Do I need to know how to code?
No. Describe a scene in ordinary words. The model authors the scene and the app renders it. Developers can also submit render jobs through the HTTP API.
Is this video compression or a video codec?
No. Coded video here means animation authored as code or a structured motion document. It does not mean a new compression format.
Can every kind of video be made this way?
No. Coded scenes work best for geometry and planned motion. Photorealistic people, live-action performances and complex physical interactions can be better suited to other tools.
Can I edit the source?
The 3D source scene is saved and available alongside your clip. This is not a full visual scene editor, and changing the source requires code knowledge or another authoring tool.